Elastic And Petrophysical Parameters Inversion Using Joint Dictionary

Seismic signal is the comprehensive response of elastic and petrophysical parameter information of underground reservoir.The inversion of reservoir elasticity and petrophysical properties by using seismic signals and logging information is an important reference for reservoir description.The traditional inversion method of elastic and petrophysical parameter information(First invert the elastic parameter information,and then invert the petrophysical parameter information by seeking the relationship between the elastic and petrophysical parameter)exists:(1)The reliability of the inversion result is low due to the step-by-step process;(2)The utilization rate of logging information is low,and the construction of the characteristic information between different parameters is incomplete,which can not accurately describe the nonlinear mapping relationship between parameters;(3)The identification method of lithology information mainly depends on the selection of parameters and discriminant function,and there is no joint information between logging parameters and lithology.Most of the existing methods are based on model driven(by establishing empirical formulas or simplifying geological models)to improve the above problems through prior information constraints,but there are still some shortcomings such as poor adaptability to geological conditions,limitations in model construction,and inaccurate description of the relationship between parameter information.In this thesis,a new seismic signal inversion method based on joint features of logging reservoir parameters and logging lithology information is proposed,which is jointly driven by model and data,and has achieved good results in model and actual data.The main innovative research work of this thesis includes the following aspects:(1)The reliability of step-by-step inversion of reservoir parameter information is low,and it can not achieve satisfactory results in the case of complex geological structures.In this thesis,a reservoir parameter information inversion method based on joint feature construction and sparse representation is proposed to obtain a complete joint dictionary set of reservoir parameter information,and then take it as a sparse representation constraint to realize the synchronous solution of reservoir elasticity and petrophysical parameter information under the framework of inversion theory.The application of model and real data shows that this method can realize the synchronous inversion of elastic and petrophysical parameters,and significantly reduce the error of step-by-step signal inversion method.(2)The construction of complex nonlinear relationship between different reservoir parameter information is incomplete due to the sparse logging information.In this thesis,a reservoir parameter information inversion method based on online joint feature construction and sparse representation is proposed.Based on the dynamic updating of dictionary information data,dictionary learning has stronger learning and representation ability,improves the accuracy of describing the complex nonlinear relationship between different parameter information,and improves the reliability of inversion results.(3)The current dictionary learning lacks a clear understanding of the meaning of information.This thesis presents a method of reservoir lithology classification inversion based on the combination of discriminant dictionary atom and logging lithology information.The model and actual data show that dictionary atoms can indeed have the significance of logging lithology,and realize the synchronous inversion of reservoir parameters and lithology information.In summary,this thesis provides a new idea for synchronous inversion of reservoir parameters and lithology information,and also provides a reference for the field of geophysical inversion,which has certain theoretical reference and practical application value.